A regression analysis is used to study the 5 most important predictors of sexual aggression, namely, sadism, alcoholism, brain pathology, history of violence, and serum levels of testosterone. Sadism will be measured by penile reactions to erotic aggresive stimuli and a score, the rape index will be derived. Alcoholism will be measured by the MAST and an alcohol erotic reactivity index derived from penile reactions after alcohol has been administered. Brain pathology will be assessed by the CT-Scan and Reitan Battery. Violence will be assessed by questionnaire and criminal record. Testosterone level will be determined by a triple serum assay. Each factor will be examined in prediction of dangerousness defined in terms of reliable psychiatric ratings of likelihood of future sexual aggression and the extent to which life threatening force will be used. Trial outcome will also be examined as a criterion variable. Five substudies are used to compare the above variables in sexually aggressive patients vs. control subjects. The latter is necessary because there are extremely few controlled studies available in the professional literature. It is expected that each predictor will significantly discriminate sexual aggression from controls. This study is unique in allowing us to determine the relative predictability of each variable in reference to each other and to sexual aggression. It is expected to improve our assessment procedures for the courts and the clinician working in a Forensic setting.